Supporting InkML in SVG2, was: Re: Agenda, 15 March 2012 SVG WG telcon

On 3/15/2012 2:36 PM, Cameron McCormack wrote:
>      [16] 
>      [17]
>    CC: reading the email again, I wonder if we didn't have an
>    action to ask charles about specific features/changes
>    … the email conclusion says "it would be fantastic if SVG could
>    be used to display InkML trace groups"
>    … but it doesn't propose anything

The replicate proposal has merit.

>    [18]
>    ml
>      [18]
>    CM: two requirements I can tease out of this
>    … one is buffered-rendering
>    … the other is like variable stroke width, but not necessarily
>    varying just stroke width, also opacity

The replicate proposal doesn't have to work with every property; it's 
limited to animate properties, it could be limited much further.

>      [20] 
>      [21] 
>    CC: we said we would not accept replicate without use cases and
>    author/implementor interest
>    … so this seems like a use case
>    … and author interest
>    … but we still don't have implementor interest
>    ED: I would want more details
> <krit> [22]
>      [22]
>    … it's not clear how the InkML would map into SVG
>    CM: the InkML traces are sequences of points with a pressure
>    value

Traces cover sensor data. You can serialize arbitrary sensor data.

Sensor API data gets serialized into InkML.

For mapping of trace groups, I'm looking at the replicate proposal.
It wasn't designed for trace groups nor was it designed with the 
shorthand of CSS transitions in mind.

Replicate proposes to use transition semantics for a repeat method. It 
wasn't designed for sensor APIs, so none of the demos feature pen results.

This is somewhat close:
David has simplified the path into a minimal set of curves.

In most cases when drawing with canvas, we take a defined shape, such as 
a circle with a radial gradient then we draw that image, interpolating 
between the points so there's sufficient overlap.

>    ED: I'm unclear how you would use it together with SVG
>    … somehow to connect the two
>    CM: whether it's a new element?
>    ED: or maybe a DOM API
>    … but the request isn't clear on that
>    DS: maybe we should ask for more clarification
>    … InkML is a big spec to require

I don't want to require the full InkML spec.

If there's a mapping mechanism, it'd be handled through a selector function.

>    CC: his derived requirements are efficient DOM rendering, and
>    variable stroke width
>    CM: but not varying opacity?

I'm looking for efficient DOM rendering and storage of sensor data, 
specifically for pressure sensitive pens.

The trace group paths offer efficient storage, so I've brought them up. 
They don't mean anything without a definition.

Pen input is usually visualized with variable transforms, not opacity.

>    RESOLUTION: SVG 2 should be able to display InkML trace groups
>    by some means, perhaps by using buffered-rendering and variable
>    stroke width, not necessarily varying anything

I'll propose something to the list using CSS semantics. The replicate 
proposal was based on SMIL, and may be a bit outdated.
It's a lot like css3 transitions with buffered-rendering turned on and 
interpolation on arbitrary values instead of just time.

CSS3 Transitions and Animations:

SMIL Transitions and Animations

If a value can be animated, it can be incremented and redrawn to the 
screen; everyone wins with a short syntax.


Received on Saturday, 17 March 2012 00:50:07 UTC